Vatican Man Inside the Anthropic

Catholic anthropologists had something to say about the recent Anthropic revolution Claude’s ruleswhich sets the behavioral parameters of the company’s AI model. Olah sent the draft to the San Jose group. The pastor, McGuire, also sent a 28-page commentary that, in his description, was less a critique of technology than “wisdom from mystics in the dark ages, from the perspective of the tension between knowing and not knowing.” Both Green and McGuire are credited for the official statement.

Undoubtedly, the conversation led Olah to find out who secretly planned to publish Leo’s book. (I wasn’t able to talk to Olah this week and I don’t know exactly how the invitation came about.) In one sense it was a risky decision. Some people who found Leo’s words inspiring were upset that he asked an industry representative to speak. Meanwhile, AI athletes felt that Olah had betrayed the AI ​​world by agreeing to a document that called for AI developers to take a break.

But the Pope had a good reason to mention Olah. An Anthropic employee has publicly voiced concerns among AI workers. They are difficult listeners to Leo’s message.

The Division of Life

The two men were not exactly compatible. In his speech, Olah talked about the secret of how AI works. He added that the models are “more abstract, more exotic, and more beautiful than science fiction imagined. They are not the computing robots we were promised. They were made from us, from our words…”

The comment seems to come down to the idea that AI species will one day be like humans. Anthropic even has an engineer dedicated to Claude’s good fortune. Leo, in paragraph 99 of his encyclical book, seems to emphasize this thinking: “We must avoid the wrong idea of ​​equating such ‘intellect’ with human beings,” he writes. He takes special pains to attack the concept of transhumanism, which he defines as the pursuit of “hybrid human machines.”

If even smart thinkers like Olah are pushing AI to rule itself—not to mention the millions of people who already consider AI models as friends or lovers—Papa Leo may face an uphill climb at this point. In conversation with Father McGuire (who uses Claude in his preparations familiesamong other things), he admitted that his nature is strange. “He is not a person, and he is not an ordinary weapon,” he says. “Nobody is saying it has a soul, but the word I’m following is that it’s a group, which we don’t know yet.”

That conflict will not be resolved for a while. The ethical questions surrounding AI development need attention now. With his partner in Anthropic, the American pope has provided the basis for a difficult discussion-if the masters of AI can stop their IPO campaigns long enough to participate.

This is a copy of Steven Levy Backchannel Newsletter. Read previous articles Here.